Retrospective Theses and Dissertations Iowa State University Capstones, Theses andDissertations

2008

Phased array system designShannon WannerIowa State University

Follow this and additional works at: https://lib.dr.iastate.edu/rtd

Part of the Electrical and Electronics Commons

This Dissertation is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State UniversityDigital Repository. It has been accepted for inclusion in Retrospective Theses and Dissertations by an authorized administrator of Iowa State UniversityDigital Repository. For more information, please contact digirep@iastate.edu.

Recommended CitationWanner, Shannon, "Phased array system design" (2008). Retrospective Theses and Dissertations. 15786.https://lib.dr.iastate.edu/rtd/15786

Phased array system design

by

Shannon Wanner

A dissertation submitted to the graduate faculty

in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Major: Electrical Engineering

Program of Study Committee: Robert J. Weber, Major Professor

Jiming Song Degang Chen Mani Mina

Stephen Vardermann

Iowa State University

Ames, Iowa

2008

Copyright © Shannon Wanner, 2008. All rights reserved.

UMI Number: 3383372

INFORMATION TO USERS

The quality of this reproduction is dependent upon the quality of the copy

submitted. Broken or indistinct print, colored or poor quality illustrations and

photographs, print bleed-through, substandard margins, and improper

alignment can adversely affect reproduction.

In the unlikely event that the author did not send a complete manuscript

and there are missing pages, these will be noted. Also, if unauthorized

copyright material had to be removed, a note will indicate the deletion.

______________________________________________________________

UMI Microform 3383372 Copyright 2009 by ProQuest LLC

All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.

_______________________________________________________________

ProQuest LLC 789 East Eisenhower Parkway

P.O. Box 1346 Ann Arbor, MI 48106-1346

ii

DEDICATIONS

I would like to dedicate this work to my parents for their enduring support and my

mother for putting up with my intolerable questions when I was younger. I would like to

thank my sister for encouraging me to take the path that is less traveled, the lord and my

savior for guiding and protecting me on this path and also dedicate my work to my professor,

Robert Weber.

iii

Table of Contents

GLOSSARY OF TERMS vii

ABSTRACT viii

CHAPTER 1. BACKGROUND 1

1.1 Introduction 1

1.1.1 Hypothesis 5

1.2 Approach 5

1.3 Summary 6

CHAPTER 2. REVIEW OF LITERATURE 7

2.1 Introduction to Phase Shifters 7

2.1.1 Time Delay 7

2.1.2 Active Phase Shift 8

2.2 Introduction to Antenna, Arrays, and Mutual Coupling 9

2.2.1 Antenna 9

2.2.2 Array Design 10

2.2.3 Introduction to Mutual Coupling 14

2.2.4 Mutual Coupling Reduction 16

2.3 Phased locked loops, Auto-tuners, and Adjustable Amplifiers 17

2.3.1 Phased Locked Loop and Acquisition Aids. 18

2.3.2 Adjustable Amplifier 18

iv

2.3.3 Automatic Tuner 19

2.3.4 Polarization Adjustment 20

2.4 Summary 21

CHAPTER 3. HARDWARE THEORY AND DESIGN METHODOLOGY 22

3.1 Introduction 22

3.2 Phase Shifters 23

3.2.1Time Delay 23

3.3 Phased locked loop Components 26

3.3.1 Adjustable Amplifier 26

3.3.2 Offset PLL 27

3.3.3 Phase Aid 30

3.4 Antennas 31

3.4.1 Antenna 31

3.4.2 Horn Antenna 34

3.4.3 Array Design 36

3.4.4 Mutual Parameters 37

3.4.5 Corrective Network and Scan Blindness 37 

3.4.6 Automatic Tuner 37

3.4.7 Modulation Format 42

3.4.8 Polarization Adjustment 44

3.5 Summary 51

v

CHAPTER 4 EXPERIMENTAL SETUP 53

4.1 Introduction 53

4.2 Setup 53

4.3 Summary 62

CHAPTER 5 EXPERIMENTAL AND SIMULATION RESULTS 63

5.1 Introduction 63

5.2 Phase Shifter 63

5.2.1 Time Delay 63

5.3 Phased Locked Loops 65

5.3.1 Offset Mixers 65

5.4 Antenna and Arrays 66

5.4.1 Antenna 66

5.4.2 Array Design 68

5.4.3 Automatic Tuner 71

5.4.4 Polarization Adjustment 72

5.4.5 Mutual Coupling 74

5.4.6 Mutual Coupling Optimization 76

5.5 Summary 76

vi

CHAPTER 6 FUTURE WORK, SUMMARY, AND DISCUSSION 78

6.1 Introduction 78

6.2 Overview of Successes 78

6.3 Future Work 79

6.4 Conclusion 79

BIBLIOGRAPHY 80

ACKNOWLEDGEMENTS 86

vii

GLOSSARY OF TERMS

CPU Computational programming unit

DSP Digital signal processing

EBG Electromagnetic band gap

FPGA Field programmable gate array

FDTD Finite difference and time domain

GL Grating lobe

HFSS A commercial finite element computational program

MIMO Multiple inputs and multiple output communication system

PA Power amplifier

PCB Printed circuit board

PD Phase detector

PLL Phased locked loop

LAN Local area network

LOS Line-of-sight-path

NLOS Non-line-of-sight-path

RF Radio frequency

Sij Scattering parameter from port i to port j

SNR Signal-to-noise ratio

QAM Quadrature amplitude modulation

QPSK Quadrature phase shift keying

VCO Voltage controlled oscillator

viii

ABSTRACT

This dissertation presents a new architecture for controlling the amplitude and phases

of signals in a fully integrated phased array system. The phase shifting is performed at

baseband and is up-converted to a RF band using a phased locked loop as a frequency

synthesizer which allows for precise control of the RF phases. A structure for a phase aid is

presented for a fast phase hopping system. A mathematical model was shown for an

adjustable amplification within a phased locked loop to mitigate phase distortion. A mutual

coupling correctional network was proposed to mitigate pattern distortion induced from the

active impedance model. An automatic tuner to reduce manufacturing variations within an

array was proposed that will reduce unwanted effects from the active impedance model and

minimize the losses due to reflections. A polarization alignment system was also proposed

that will help improve the overall gain between the transmitting and receiving system.

1

CHAPTER 1. BACKGROUND

1.1 Introduction

The phased array antenna system has

many applications in wireless applications,

especially MIMO (multiple inputs and

multiple outputs) communication. By using

multiple antennas to transmit and receive

the signal, the transmit rate is pushed closer

towards the channel capacity limit while simultaneously improving security [2]. Another

application of such a system is in Sensor Array Networks where information from a single

sensor can be directed or transmitted to a specific receiver by steering the antenna in the right

direction [3]. Since the transmitted signal can be steered to a specific receiver and nulled in

other directions, the security of the signal can be improved. A phased array antenna system

can be utilized by the military to transmit and receive secure information [2]. It also has

applications in mobile LANs, adaptive dynamic array processing for antennas and

automotive radars for collision control, path/lane control, etc [4], [5]. Phased array

applications have been around since the early 1940’s when first developed with military

intentions. The primary purpose of a phased array is to agilely steer the main beam of an

antenna in a given angular direction of the line-of-sight-path (LOS) which is labeled as the

direct path in Fig. 1.1. This will enhance the reception of the direct path signal and reduce the

reception of non-line-of-sight (NLOS) components which is labeled as the reflected paths in

Figure 1.1: Illustration of Multi-path fading in a communication link [1]

2

Fig 1.1. As a result of steering, the multipath effects of the communication link will be

reduced. The reduction of multipath fading will inherently increase the channel capacity of

the communication link. A demonstration of steering by a phased array is illustrated in Fig

1.2. This is accomplished by setting phases and amplitudes on each element of the array. A

secondary aspect of phased array antennas is the ability to add a null in the direction of an

unwanted signal, for instance an enemy trying to jam the communication system or a

Figure 1.2: The steering of the main beam of antenna at 0, 90, and 180 degrees for part A, B, and C respectively. The simulation was done using Ardtool for Matlab with a 10 element, omni-directional, ordinary end fire array spaced a half wavelength apart [6].

3

hacker trying to gain access to a private network [2]. A majority of the work done in the early

development of phased array has been performed in the military settings thus limiting the

commercial development of the area. The primary cause of this limitation to military rather

than the commercial environment was due to the cost and accurate characterization of the

components. Today’s technology has matured significantly enough to allow for the

technology to break into the commercial world. The accuracy of simulations has been greatly

improved through both commercial and proprietary computational electromagnetic programs.

The limitations of these programs are primarily due to the requirement of a high-end CPU to

Figure 1.3: Part A. shows the response of a filter that corresponds to an equal amplitude phased array, part B shows the response of a filter that corresponds to a binomial weighted phased array and part C shows the response of a Tscebyscheff weighted phased array. The simulation was done using Ardtool for Matlab with a 10 element, omni-directional, ordinary end fire array [6].

4

produce an accurate simulation which increases with the size of the grid. Integration of the

RF components onto a silicon chip has reduced the variability of the amplitude and phase

characteristics of the devices and has significantly driven the cost down. Beam formation

algorithms have in the past been developed using spatial filter theory and these have been

thoroughly studied. Several responses of different spatial filters are shown in Fig. 1.3.

Similar tradeoffs between stop band and pass band characteristics in a circuit filter design can

be observed in an array such as shown in Fig 1.3. There will be a tradeoff between directivity

Figure 1.4: Results of optimization routines for beam formation in the direction of line sight for incoming signals at 6/πθ = and 3/2πθ = with 4 omni-directional elements spaced 0.075 m apart on the x-axis at 2 GHz. Part A shows the results of the ESPIRIT algorithm, part B shows the result of the Temporal Reference algorithm, and Part C shows the result of the MUSIC algorithm using the Smart Antenna System for Mobile Communication [7].

5

in the pass band and side lobe ripples in the stop band. An inherent drawback of this theory is

accurate adjustability of the phase and an amplitude characteristic is required for each

elemental excitation of the phased array. It is also extremely complex to produce an optimal

design in a large phased array in a non-static environment. Current research in beam

formation is producing designs that adaptively converge to an optimal solution using

optimization techniques such as convex, genetic, linear, and gradient optimization

algorithms. Responses from MUSIC, Temporal Reference, ESPIRIT algorithms are shown in

Fig. 1.4. There will be a tradeoff between the ability of the algorithm to identify the direction

of the incoming signal within a pass band and rejection of other signals within the stop band.

This can be observed in part A and part B of Fig 1.4. The algorithm in Part C. of Fig 1.4 is

only used to identify the direction of the incoming signals. This information can then be

utilized to design an appropriate filter.

1.1.1 Hypothesis

An integrated phased array system can be built using shift registers and phase locked

loops, to control the phase of the electronics with arbitrary accuracy. This will in turn allow

the direction of the beam pattern of the main beam of the array to be digitally controlled and

allow nulls to be set in a predictable fashion.

1.2 Approach

A three by three phased array was built and tested. An FPGA was used to perform the

digital circuitry tasks and a phased locked loop was used as an RF up-converter to control the

main beam of the antenna in a predictable manner. The system was measured in an anechoic

6

chamber in order to avoid interference, reflections, and to allow for precise measurement of

the beam pattern in the far field measurement location.

1.3 Summary

The hardware platform, which was outlined in Sec. 1.2, would enhance the current

state of the art by allowing more accurate phased array systems to be built with less expense.

This in turn allows the field of space-time adaptive communications and MIMO systems to

break the barrier of moving beyond theoretical ideas into product development. The need for

such systems will become more critical as demands for wireless bandwidth increases

worldwide. Better control of beam patterns is already becoming critical in densely populated

areas such as European countries.

7

.delayrfshift Tωθ =

.delayrfshift Tωθ =

CHAPTER 2. REVIEW OF LITERATURE

2.1 Introduction to Phase Shifters

A phase adjusting circuit (phase shifter, ect.) is absolutely critical in a phased array

system. The evaluation of phase shifters is most often evaluated in terms of cost, linearity,

and available phase shift. Phase shifters can be broken down into three different categories:

time delay element, filter element, or active devices.

2.1.1 Time Delay

Time delay elements have often been reported using microstrip lines [8]. An inherent

drawback of a time delay element is that most often they are a non-variable phase shifter

unless switches are implemented and then the number of different phase shifts available is

directly related to the number of switches, thus increasing the overall cost of the system.

Another drawback of using transmission lines as a phase shifter is that the size is inversely

proportional to frequency and most commercial applications are implemented below 10 GHz

where transmission lines are physically large. The shift in time, delayT , of a transmission line is

related to the phase and the frequency of operation, rfω , as shown below

The resultant output of the shift can then be represented as

)1.2(

)2.2(

8

).sin( shiftrfinout tVV θω +=

( )( ) ,

)()(tan 1⎟⎟⎠

⎞⎜⎜⎝

⎛= −

ωωθ

HrealHimag

shift

Phase shifters implemented using filters are most often done using a variable reactance

element such as a varactor [9], or ferrite based device [10], thus making the transfer function

variable. These have been implemented at baseband, RF, and directly at the antenna’s

terminals. In most cases, the inherent drawback of variable phase shifters lies in the drift of

the phase shift and the linearity of the phase shifter. The phase shift of a filter can be

determined by its phase transfer function,

where ( )ωH is the transfer function of the filter. The magnitude of the transfer function will

also vary when adjusting the reactive elements of the filter. This will cause a coupling

between the amplitude and phase of the excitation and produce a distortion in the desired

array pattern as the filter is varied to steer the array.

2.1.2 Active Phase Shift

Active phase shifters have been produced by using injection locking properties of

voltage controlled oscillators [11] or phased locked loops [12]. In the past, the inherent

drawback of phased locked loops has been the cost of their implementation. This has been

remedied, however, by the decreasing cost of integrated phased locked loops and the

increasing availability of phased locked loops that operate above 1 GHz. This availability of

phased locked loops above 1 GHz will likely continue until Moore’s Law ceases to be true.

An inherent gain of the phased locked loop implementation is the ability to perform a base

)4.2(

)3.2(

9

band phase shift. That is typically performed at a lower cost solution than with a RF shifter.

The inherent drawback of injection locking is that the signals on the antenna elements are

coupled to their nearest neighbors and this reduces the degrees of freedom allowed for beam

formation although it is sufficient enough for crude scanning.

2.2 Introduction to Antenna, Arrays, and Mutual Coupling

Antennas have progressed from simple radiating elements, such as a simple monopole

or a loop antenna, to a more integrated approach such as using a microstrip antenna or an

inverted F antenna. This integration allows the feasibility of building large arrays with super-

directivity properties; however, a large inherent drawback of large arrays is increased mutual

coupling that will distort the field pattern [13]. The increase in mutual coupling has led to an

increase in methods to deal with or reduce the mutual coupling between elements.

2.2.1 Antenna

In modern day consumer electronics, there are four different types of antennas in use.

These are the inverted F, the microstrip antenna, the helix antenna and the dipole antenna.

The selection criteria for any antenna can be reduced down to its polarization,

manufacturability, bandwidth, and field pattern. The manufacturability considerations

include cost of material, robustness against manufacturing errors, mutual coupling in the

array, and integration of the components. Field pattern considerations can be broken down

into two categories - bandwidth and directivity of the antenna. The application dictates which

property is more desirable. Dipole antenna’s advantages are H-plane omni-directional field

patterns, a medium wide bandwidth, and medium manufacturing complexity [14]. A

10

microstrip antenna’s advantages are ease of manufacturing, dual polarization capabilities, and

ultra-wide bandwidth capabilities [14]. An inverted F antenna’s advantages are an

approximate omni-directional pattern, ease of manufacturing, and medium width bandwidth

capabilities. A helical antenna’s advantages are an H-plane omni-directional field pattern

capability, a dual polarization capability, and a high bandwidth [14]. Typical criteria for an

antenna in a phased array system are directivity, dual polarization support, and ease of

manufacturability. Therefore, the microstrip antenna often seems to be the most suitable

antenna. There are numerous variants of microstrip antennas. The selection criterion depends

on polarization capability and gain.

2.2.2 Array Design

Previous work [14], not including mutual coupling effects, suggest that any antenna

array can be factored into two components. One component, elementE , contains the field pattern

of a single radiating element and the second component, AF , that contains geometric

information concerning how the antennas are arranged [14]

The importance of this separation of terms is the fact that the arrays pass band and stop band

can be designed by controlling both the amplitudes and phases of the array factor. The

element field pattern then only affects the angle at which the pass band can be placed and the

directivity of the overall array pattern. Consider an array with M elements in the x-axis

direction and N elements in the y-axis direction, assuming then that the antennas are

.1

AFEEE element

N

n

nelementfieldtotal == ∑

=− )5.2(

11

positioned linearly about the x-axis and y-axis, and M and N are odd, the array factor can

then be defined as [14]

where d is the center-to-center spacing of the elements. The array factor is solely

determined by the amplitude, nmA , and phase, nmθ , of the excitation placed on each element.

The array factors for other geometries, such as circular, rectangular, etc., arrays have also

been formulated [16] [17]. An excitation can be defined as the following

.ˆonmjkr

nmnm eAA ψ⋅−=

Figure 2.1: Coordinate system for derivations for steering the array beam [15].

( ) ( ) ,sin)sin(2/)1(

2/)1(

2/)1(

2/)1(

cos)sin( φθθφθ jkdnM

Mm

jN

Nnnm

jkdm eeAeAF nm∑ ∑−

−−=

−−

−−=

= )6.2(

)7.2(

12

.sinsind

noGL

λθθ +=

.sin11

o

dθλ +

=

The position vector, nmr , is dependent upon the location of the patch, which is defined in Fig

2.1. The main beam of the antenna can be steered in the direction of ( )oo φθ , by setting the

phases on the antenna element where the resulting variable oψ̂ is given by [17]

The vector, oψ̂ , is defined from the center of the array to center of the main lobe. This

results in a constant beam direction due to the fact that the corresponding exponential terms

in the array factor are canceled [15]. One concern in the design of a phased array is the

appearance of grating lobes. A grating lobe is the duplication of the main lobe at a different

angle. An example of a grating lobe can be observed in Part B of Fig 1.2. Gating lobes can

appear under the following conditions

where n is an integer. The grating lobe appears at an angle, GLθ , determined by [15]

with

Without an appearance of a grating lobe, for a given scan angle oθ at a frequency f , the

maximum element spacing is given by [15]

Typical element spacing for patch array antennas is given by [15]

( ) ,2sinsin2 ndo πθθ

λπ =−

.ˆcosˆsinsinˆcossinˆ zoyooxooo aaa θφθφθψ ++=

.5.0 λλ ≤≤ d

)8.2(

)9.2(

)10.2(

)11.2(

)12.2(

)13.2(

.1sin ≤GLθ

13

Coupling versus Distance

-45.5

-45

-44.5

-44

-43.5

-43

-42.5

-42

-41.50 0.5 1 1.5 2

lambda/d

S21

Rel. dielectric=2

The closer the antennas are to each other, the more the mutual coupling increases.

The further the antennas are apart the more that grating lobes are introduced. The appearance

of grating lobes is similar to aliasing of a sampled signal as derived in signal processing

theory. The placement of an antenna is effectively sampling in the spatial domain. In signal

processing, one must sample the signal at a rate of twice the highest frequency component in

order to avoid aliasing of the signal. Similarly, one must sample the spatial domain at a rate

that is twice the wavelength of operation in order to avoid the appearance of grating lobes.

The relationship between distance and mutual coupling is shown in Fig. 2.2. Mutual coupling

in dB’s can be seen as a function of distance, relative dielectric, height of PCB, and

frequency of operation [18].

Fig 2.2: Mutual coupling coefficients in dB versus spacing in a phased array at 11 GHz [18].

14

Fig 2.4: Flow diagram of an antenna array

2.2.3 Introduction to Mutual Coupling

The most commonly used method for calculating mutual coupling is through the

spectral domain approach. This

methodology, first proposed by Pozar,

accounts for spatial, reflective, and

surface wave coupling [19-20]. These

coupling waves are as shown in Fig.

2.3. The type of coupling that is prevalent determines the hardware mechanism that is

available to reduce the coupling. Pozar introduced the idea of the active impedance which

states that the input impedance of one antenna element is affected by the coupling of the

output of another antenna element [13]. The effects in a phased array design were explored

by Dodov, who found that surface wave and reflective coupling are only prominent in long

arrays and also stated the conditions for excitation of such waves [18]. The Floquet model

was introduced to predict scan blindness in an infinite array [21].

Fig 2.3: A diagram showing the various methods that a field emanated from one antenna element interacts with neighboring antenna elements [18].

15

,})(

){(sin)2/(

2112222

sin)2/(1221111

φ

φφ

kdj

kdjelement

eSASAA

eSASAAEE−+++

++∝

In a coherent system, the reflection coefficient is dependent upon the neighboring

antennas. Assuming coherency, mutual coupling effects, to a first order approximation, can

be described in terms of an active reflection coefficient as can be derived from the flow

diagram in Fig. 2.4 and given below [13]

The field pattern of a two element array shown in Fig. 2.5 can be described in terms of

mutual coupling parameters as

,121

211

1

1221111 S

AAS

ASASA

VV

trans

refl +=+

==Γ

.212

122

2

2112222 S

AAS

ASASA

VV

trans

refl +=+

==Γ

Fig. 2.5: A two element array with a microstrip array (Left) and dipole array (Right).

)14.2(

)15.2(

)16.2(

16

where 1A and 2A are the complex feed excitations and elementE is the field pattern emitted by

a single antenna.

2.2.4 Mutual Coupling Reduction

There have been two different methods used to reduce mutual coupling effects in arrays,

electromagnetic band gap structures [23] and perforated ground planes [24]. Electromagnetic

band gap structures (EBG) and a perforated ground plane shown in Fig. 2.6, allow for the

physical reduction of coupling due to surface waves by removing the electromagnetic

boundary conditions which allow them to propagate. These techniques have been shown to

reduce mutual coupling up to -15 dB. The disadvantage of an EBG and a perforated ground

plane is that they do not reduce coupling through the air or through reflection and therefore

must be combined with the predistortion technique in order to get an undistorted field. A

disadvantage of pre-distortion techniques is in the fact that the technique must be adaptive

due to environmental changes that will affect coupling parameters. The EBG and the

perforated ground plane are passive structures.

Fig 2.6: A perforated ground plane (Left) and an electromagnetic band gap (Right) [24], [23].

17

2.3 Phased locked loops, Auto-tuners, and Adjustable Amplifiers

Phased locked loops have progressed from a purely analog form to a mostly digital

approach. The integration of components in a phased locked loop has almost reached full

potential. The filter is still left external from the VLSI chip to allow for adaptability of the

phased locked loop to a wide variety of applications. Auto-tuners have been recently

proposed in the last couple of decades [28]. The reason that they are not being implemented

is primarily due to the non-linear nature of a varactor or pin diode. These devices will

generate harmonic content that couples into the antenna when large output signals are

induced from a power amplifier. Adjustable gain amplifiers have been developed using pin

diodes as attenuators due to the low cost of components and ease of implementation.

However, amplifiers using adjustable gain do exist and utilize current versus gain more

efficiently. The gain of the amplifier can be written as

Type 1 Filter Type 2 Filter Type 3 Filter Phase position Zero Zero Zero Phase velocity Constant Zero Zero phase acceleration Increasing Constant Zero

outm RgGain =

Table 2.1: Phase error resulting from different types of phase inputs [26].

)17.2(

18

in terms of it’s input transconductance, mg , and the output impedance, outR , where mg is

dependent on the device current. Therefore, the gain can be made variable by either adjusting

the load impedance or bias current.

2.3.1 Phased Locked Loop and Acquisition Aids.

The acquisition aids published to date include but are not too limited to an adjustable

filter in the phased locked loop. The acquisition time required to lock in a signal can be found

to be proportional to the bandwidth of the input filter. However, larger bandwidths will

cause additional phase noise and spurious content to appear in the output signal. Therefore,

Lui and Li purposed adjusting the bandwidths of the filters based on the magnitude of the

error signal [25], thus, incorporating the best properties of high bandwidth and low

bandwidth in the phased locked loop. The type of filter used depends on the type of input

signal and the phase error that can be tolerated in the output signal, which can be determined

from Table 2.1. Typical application of type 1 or type 2 filters would be in a frequency

synthesizer where phase acceleration is kept at a minimum and a type 3 filter could be used

in a frequency recovery system where phase acceleration is a major concern. There has been

considerable research in the architecture of all three types of filters that will guarantee a

stable system [27].

2.3.2 Adjustable Amplifier

Typically, even though adjustable gain provides higher degrees of freedom in terms

of beam formation, its use in a phased array is limited due to the fact that adjusting the gain

19

of the amplifier or using a variable attenuator, will change the phase characteristics on each

radiator. This effect can be minimized and be de-embedded from the system by using a look

up table and through the use of pre-distorting.

2.3.3 Automatic Tuner

Previous work in developing automatic tuning systems for antennas consists primarily of one

structure as shown in Fig. 2.7 [28]. The reflected energy from the antenna is sensed by a

power detector through a coupler. The reflected energy can then be used as an objective

function in conjunction with an optimization algorithm to tune the antenna. Advantages of

the structure consist of minimizing loss and beam distortion within the system.

Disadvantages include the harmonics induced due to a large signal characteristic at the output

of the PA in the transmitting mode. These effects are less detrimental in a receiving mode.

However, De Mingo, Valdovinos, Crespo, Navarro, and Garcia suggested using notch filters

to deal with harmonic generation as has been similarly done for power amplifiers [28].

These authors investigated to a limited extent the properties of adaptive, deterministic, and

hybrid tuning algorithms.

Fig 2.7: An automatic tuner for controlling the manufacturing mismatches in the antenna [28].

20

2.3.4 Polarization Adjustment

There have been numerous works describing microstrip patches that can radiate in

different kinds of polarization modes. However, only one paper was found that describes a

method of changing antenna polarization deterministically. The approach is shown in Fig.

2.8. The advantages of such a structure are that the polarization of transmitting or receiving

signals can be tuned to the incoming wave and thus maximizing reception. The disadvantage

is that the diodes go non-linear due to large signal characteristics of the signal at the output of

a transmitter. This is less detrimental in the receiving mode where the alignment can be

more easily performed. The tuning of the polarization is performed through the use of a

varactor. The varactor capacitance can be change to dictate whether the length or the width of

the antenna is resonant. This in turn determines the polarization of the antenna.

Fig 2.8: Capture range of an automatic tuner for various mismatch resonances for different types of algorithms [29]

21

2.4 Summary

All previous results were judged on implementation, cost effectiveness, and

manufacturability following current and future industrial trends. The author acknowledges

that future trends may be subject to change based on innovation but remains firm on his

criteria. The antenna array design and exploration has been explored. Future work needs to

be implemented in randomization algorithms for the antenna design. Mutual coupling

reduction methods have been explored but improvements are needed to determine which

methods could be combined with baseband pre-distortion to further mitigate the effects.

Phased locked loop aids do exist but need improvement if a fast phase hopping system is to

be implemented with increasing data rates. Adjustable amplifiers also exist but their phase

distortion will cause undesirable effects in the phased array. Further reduction of coupling

will always be a benefit. The area of automatic tuners has been explored but has not been

widely utilized. More experimental data, algorithms, and hardware design in the presence of

a large signal inputs and outputs need to be obtained. Automatic tuners may become outdated

if broadband antennas are capable of being designed with randomization procedures but may

still be utilized to fine tune broadband antennas. Polarization adjustments have been explored

but not utilized. The current approach for dealing with mismatch of polarization is by using

multiple antennas with varying polarization. The author believes that an automatic

polarization adjustment may be a more optimal in terms of both performance and cost. In

summary, there is room for enhancement and continuing work in the field of phased array

systems.

22

CHAPTER 3. HARDWARE THEORY

AND DESIGN METHODOLOGY

3.1 Introduction

The phase shifter section describes a shift register to produce fixed time delays of the

input signal and forms the original basis for this project. The time delay values can then be

chosen through the use of a multiplexer to pick the appropriate delay value. A latch register

is used at the output of the multiplexer to produce a synchronous system. A phased locked

loop is then utilized to up-convert the baseband signal to RF. The phased locked loop section

describes an adjustable amplifier that is used provide more degrees of freedom to a phased

array, a phase aid to provide a fast phase hopping system, and an offset PLL to provide a

scaled free version of the hardware phase shifter. These were all added to the project to deal

with specific problems and to provide a more optimal design. The antenna section describes a

horn antenna design, which was used in experimentation, as well as a microstrip antenna and

design procedures for an array. In addition, the antenna section also describes mutual

coupling effects with a corrective network, scan blindness effects, and a polarization

alignment scheme proposed to increase the SNR of the system.

23

3.2 Phase Shifters

3.2.1Time Delay

A single element hardware block diagram for a prototype variable phase shifter is

shown in Fig. 3.1 [30]. A reference clock is divided down by 16 to provide a data source and

the output signal is represented as

where T is the time period. The shift registers are shifted at one sixteenth of the clock rate of

the system. The shift register then contains 16 different delayed versions, sampled on the rising

edge of the clock, oω , as shown in below equation

Fig 3.1: A diagram showing the elements used in a one element phase shifter.

,)162sin(1

4),(

..5,3,1∑∞

=

+=i

oshifteditl

litV πωπ

.)2sin(14

)(..5,3,1

∑∞

=

=n

ref Tnt

ntV ππ )1.3(

)2.3(

24

for 16,..,2,1=i [30]. The PLL (phased locked loop) locks into phase with the shifted data and

provides a 2.425 GHz source that is represented as

where

and loopdelayφ is the phase delay in the feedback network for 16,..,2,1=i and where M and N are

the frequency divide ratios of the reference and the RF signal of the phased locked loop

respectively as shown in Fig. 3.2.

Fig 3.2: A diagram showing the basic functional elements within a phased locked loop.

πφπ ≤≤− i0

,162

0 loopdelayi i

MN φπφ += )3.3(

)4.3(

25

The division ratio produces a scaled version of the phase offset of the reference at the

system output. Scaling can be minimized or eliminated by using a frequency offset system.

In order to ensure stability and zero steady state phase error during phase hops, the phased

locked loop contains a third order loop filter [27]. The settling time of the phased locked loop

used can be seen in Fig. 3.3. The response determines the maximum bit rate that can be

transmitted through the communication. The signal is sent to a power amplifier whose desired

load impedance is matched to the inactive input impedance of the antenna terminals. The input

signal can be represented as

with ω being the RF output frequency and i0φ is the phase shift introduced by the PLL and

shift register.

Fig 3.3: A simulated phase error plot for the PLL unit step phase response.

),sin(),( 0,i

pantenna tCitV φω += )5.3(

26

( ) ( ) ).()(1)(1

11 s

sGHK

ssGH

s distortionn

e θθθ+

−+

=

3.3 Phased locked loop Components

3.3.1 Adjustable Amplifier

As shown in Fig. 3.4, in order to mitigate any phase error that could be caused by an

adjustable gain amplifier, the amplifier is placed within a phased locked loop that provides

negative feedback to reduce error caused by adjusting the amplitude of the signal. The

Laplace transform of the loop gain is given by

The dynamic phase error can be written as follows

Where 1θ and 2θ are shown in Fig 3.4. The error can then be further manipulated using

standard control theory in terms of an input phase change and the distortion phase change

( ) ( ) ( ),21 ssse θθθ −=

./ sKKKKG nofp=

Fig 3.4: A phased locked loop diagram which includes an adjustable gain amplifier element.

)6.3(

)7.3(

)8.3(

27

( ) ( )).(

)(11

)(1)( 1 s

sGHKs

sGHsGHs distortion

no θ

θθ

++

+=

If the filtering is designed properly, the steady state error should reduce to zero. The time

taken for convergence will then dictate the maximum bit rate that can be passed through such

a system. The output phase can be found from the input phase by the following

This derivation is similar to the standard PLL response derivation with the exception of an

additional transient term. The transient’s time constant will depend on the open loop transfer

function. This distortion correction technique is very similar to a Cartesian Feedback loop

that is used for correction of amplitude to phase modulation conversion [32]. In this project, a

pin diode attenuator was proposed for the adjustable amplifier due to its cost effectiveness

and ease of manufacturing. An adjustable gain amplifier may be more desirable due to its

higher efficiency.

3.3.2 Offset PLL

The multiplicative phase factor, caused by digital division in the feedback loop, can be

eliminated by using mixers to translate the RF signal to baseband as shown in Fig. 3.5. In

steady state, the mixers will produce a constant phase error which will be uniform over all

antenna elements. Due to the uniformity of the phase error among the antenna elements, the

error can be factored out of the array equation and thus mitigating a possibility of distortion.

)9.3(

28

The phase error can be written as

where 1θ , and 2θ are shown in Fig 3.5. The transfer function of the error can then be written

as

The output can then be related to the error by

In the steady state, the error will approach zero assuming that the order of the denominator is

of a higher power than the numerator and the filter is stable

Thus,

( ) ( ) ).()(1

1)()(1

1)(1

1211 s

sGHs

sGHs

sGHs offsetoffsete θθθθ

+−

+−

+=

( ) ( ) ( ) ( ) ( ) .0][0 211 =++−== sssss offsetoffsetoe θθθθθ

Fig 3.5: A phased locked loop using mixers as a frequency offset for an output conversion to the reference frequency.

( ) ( ) ( ) ( ) ( )].[ 211 sssss offsetoffsetoe θθθθθ ++−=

( ) ( ) ( ).21 ssse θθθ −= )10.3(

)11.3(

)12.3(

)13.3(

29

In order to ensure that no false locks occur due to harmonics, a frequency planning technique

of the following frequencies was used

where k, l, m, and n are integers and VCOω , 1offsetω , and 2offsetω are the output frequency of the

VCO, and LO frequencies for the offset mixers respectively. Adequate filtering was then

applied to the corresponding harmonics.

Fig 3.6: A diagram of a phased locked loop with an acquisition aid element inserted behind the VCO.

( ) ,21 offsetVCOoffset lnmk ωωω ++

.211 offsetoffseto θθθθ ++= )14.3(

)15.3(

30

3.3.3 Phase Aid

Phase aids can be introduced into a phased locked loop to reduce the transient time needed

for convergence. The proposed structure shown in Fig. 3.6 is used to expedite convergence

by summing a small delta phase error in the loop. The loop gain of the system shown in Fig

3.6 can be written as

The phase error can be written as

where 1θ , and 2θ are as shown in Fig 3.6. The equation can be further manipulated to

produce the following

.

The error can then be related to the output by the following equation

The additional offset term leads to an extra transient term

./ sKKKKG nofp=

( ) ( ) ( ).21 ssse θθθ −=

( ) ( ) ).())(1()(1

11 s

sGHsKK

ssGH

s offsetno

e θθθ+

−+

=

( ) ( )).(

))(1()(1)( 1 s

sGHsK

Ks

sGHsGHs offset

o

no θ

θθ

++

+=

( ) ( ) ( ).)( ss

KK

ssGHs offset

o

n

eo θ

θθ +=

)16.3(

)17.3(

)18.3(

)19.3(

)20.3(

31

This extra transient can be formed such that when in combination with the original response

it produces a pseudo convergence of the loop.

3.4 Antennas

3.4.1 Antenna

The rectangular microstrip patch was chosen due to its popularity in an integrated

transceiver, its capability of supporting dual polarizations, the ease of its manufacture, and

having a well known approximate solution for its far field pattern. An example is shown in

Fig. 3.7 [14]. The basic design of the microstrip antenna can be done in a five step process

when the microstrip is treated as a resonant cavity. The assumptions for this model consist of

enclosing the microstrip patch within a perfect magnetic conductor (PMC) artificially

extending the length of the cavity to

Fig 3.7: A diagram indicating length extension due to a fringing effect of the electric field and the corresponding dimension of a microstrip patch [34]

32

account for fringing effects, having an infinite ground plane, and assuming a dielectric, rε ,

exist only under the patch. The benefit of using the cavity model is that the main field lobe

pattern is closely approximated. The disadvantages for the model include diffraction caused

by the dielectric and finite ground plane that gives discrepancies for the field pattern for

angles near grazing and behind the ground plane [33]. The dimensions can be calculated by

the following sequence of steps.

1. Determine the [14]

2. Calculate the length extension due to fringing effects caused by the electrical field

extending past the microstrip patch as shown in Fig. 3.7. The effective dielectric constant due

to the air and dielectric combined can be approximated by [14]

and the length extensions from the fringing field can then be approximated by [14]

( )

( ).

8.0258.0

264.03.0412.0

⎟⎠⎞

⎜⎝⎛ +−

⎟⎠⎞

⎜⎝⎛ ++

=Δ

hw

hw

hLreff

reff

ε

ε

,1212

12

1 2/1−

⎥⎦⎤

⎢⎣⎡ +

−+

+=

whrr

reffεεε

.1

22

1+

=rrf

Wεεμ οο

)21.3(

)22.3(

)23.3(

33

3. Calculate the resonate length of the microstrip

The resonate length with the length extensions subtracted can then be

calculated by [14]

4. Determine the impedance and probe location

The input impedance for different probe locations was performed using FDTD

software. A probe location which will produce a 50 ohm input impedance into the

patch can be found iteratively.

5. The resulting field pattern is [14]

and

where

.22

1 Lf

Lreffr

Δ−=εεμ οο

)),cos(2

( sinc)]sin(2

[ sinc)sin( θθθπφ

WkhkreWVk

jE oorjk

ooplaneHo−

− =

.)sin()( sincx

xx =

))sin(2

cos()]cos(2

[ sinc φφπφ

effoorjk

ooplaneE LkhkreWVkjE

o−− =

)24.3(

)25.3(

)26.3(

)27.3(

34

3.4.2 Horn Antenna

The pyramidal horn is widely used for making experimental measurements since the

gain of the antenna can be made arbitrarily large as a function of it size. One example of

pyramidal horn is shown in Fig. 3.8. The gain, oG of an antenna is directly related to the

directivity, oD , of the antenna by the following formula

where oε is the loss factor of the antenna system. The large gain of the pyramidal horn

causes a narrow beam pattern for the transmitting and by reciprocity the receiving field

pattern. The narrow pattern allows for accurate measurements of transmitting and receiving

antenna field patterns in particular at their null locations. In terms of signal processing, the

use of the horn antenna measurement can be viewed as a convolution operation. The field

pattern of a narrow beam horn antenna can be approximated by a delta function and thus the

measured field pattern convolved with a delta function will return the measured field pattern.

The approximation of a delta function depends on the directivity of the horn antenna. The

gain of the antenna, under lossless conditions, is directly proportional to the physical area of

the horn antenna, and is given by [14]

where 1a and 1b are the outer dimensions of the face of horn. Assuming that the dimensions of

the horn 1a , 1b and the dimensions of the rectangular feed a ,b satisfy the following condition

[14]

).(421

112 baGo λπ

=

,ooo DG ε=

,116

123

2)12(2 3

222

⎟⎟⎠

⎞⎜⎜⎝

⎛−⎟

⎟⎠

⎞⎜⎜⎝

⎛−=−⎟

⎠⎞

⎜⎝⎛ −

χπλχππχ

λχ oo GaGb

)28.3(

)29.3(

)30.3(

35

where

and

the corresponding dimensions of the antenna can then be iteratively solved for by using the

following design technique.

1. Pick a desired value of oG , and iteratively solve for χ

2. Determine the corresponding values of eρ and hρ

3. Find the corresponding values of 1a and 1b by using the following equation [14]

and

χλρ

=e

,18 3

2

⎟⎟⎠

⎞⎜⎜⎝

⎛=

χπλρ oh G

λπχπ

λρ2

32

21o

hG

a =≈

Fig 3.8: Picture of a pyramidal horn antenna with dimensions.

.221 χλλρ =≈ eb

)31.3(

)32.3(

)33.3(

)34.3(

36

3.4.3 Array Design

The array was designed at an operating frequency of 2.424 GHz. The array consisted

of a rectangular 3 by 3 array. The size of the array was based solely on manufacturing

abilities available and the shape was chosen based on the ease of mathematical formulations.

The individual antennas were spaced half a lambda apart in order to avoid spatial aliasing,

grating lobes, and to decrease the effects of mutual coupling. With 3 elements centered along

the x-axis and 3 elements centered along the y-axis, there are a total of K=3x3 elements.

Elements can be counted by rows with a single index k=1…9, so that )3,1mod(1 −+= kik ,

and )3/)1int((1 −+= kjk [13]. Including mutual coupling effects, the electric field can be

described by the following

where elementE is the dominate polarization of the field pattern [13], A is the elemental

excitation,

and

,][),(),(9

1

])2()2[(9

1∑∑=

−+−

=

+∝m

vjuijnmm

nnelement

nneSAAEE φθφθ

φθ cossinkdu =

.sinsin φθkdv =

)35.3(

)36.3(

)37.3(

37

3.4.4 Mutual Parameters

Assuming the same coupling solution as stated in the active impedance section for a

two element array, the field pattern can be written in the following form

Using the objective function [22]

where )(φE is the measured field pattern with known excitation coefficients, one can

minimize the squared error using optimization techniques and solve for the unknown

scattering parameters 11S , 12S , 21S , and 22S [35], [36]. A uniform random search was chosen

for this work due to its ease of implementation. A genetic optimization or a gradient search

may produce more accurate results and be computationally more efficient. The minimization

should be performed at multiple scanning angles to ensure the correct solution.

3.4.5 Corrective Network and Scan Blindness

Assume that the electric field pattern can be described by the following formula

and that the desired field pattern can be described with the following coefficients

( ) ,),,,,()(),,,(2

0

22221121122211211 ∫ −=

π

φφφ dSSSSEESSSSf

,})(

){(sin)2/(

2112222

sin)2/(1221111

φ

φφ

kdj

kdjelement

coupled

eSASAA

eSASAAEE−+++

++∝

.)(

)(),,,,(

sin)2/(2112222

sin)2/(1221111

22211211 ⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

+++

++∝

− φ

φ

φkdj

kdj

element eSASAA

eSASAAESSSSE )38.3(

)39.3(

)40.3(

38

The arrangement of the array is as shown in Fig 2.5. The coefficients can then be presented

in matrix form

where 1A and 2A are the complex input excitations, and 1C and 2C are the desired complex

output excitations. The inversion can be accomplished by using Cramer’s rule. The input

excitations can then be solved for and implemented using phase shifters and adjustable

amplifiers as shown in Fig. 3.9. 1A and 2A are given by

The scattering parameters can be obtained either through direct measurement [31, pp.61-80]

through the method of open and short circuits [37, pp.124],

or through far field measurements [13]. Scan blindness, resulting in a lack of radiation or

reception in a particular direction, can occur during the following conditions

11 21 1 1 21 2 12

12 22 2 2 11 1 22

1 0and

1 0 1 1S S A A S A S

S S A A S A S+⎡ ⎤ ⎡ ⎤ ⎡ ⎤ − −

= ⇒ = =⎢ ⎥ ⎢ ⎥ ⎢ ⎥+ + +⎣ ⎦⎣ ⎦ ⎣ ⎦

or

,)1)(1(

)1(

21122211

2121221 SSSS

CSCSA−++

−+=

⎥⎦

⎤⎢⎣

⎡⎥⎦

⎤⎢⎣

⎡+

+=⎥

⎦

⎤⎢⎣

⎡⇒⎥

⎦

⎤⎢⎣

⎡⎥⎦

⎤⎢⎣

⎡+

+=⎥

⎦

⎤⎢⎣

⎡−

2

11

2221

1211

2

1

2

1

2221

1211

2

1

11

11

CC

SSSS

AA

AA

SSSS

CC

}.{ sin)2/(2

sin)2/(1

φφφ

kdjkdjelement

desired eCeCEE −+∝

.)1)(1(

)1(

21122211

1212112 SSSS

CSCSA−++

−+=

.)1)(1( 21121122 SSSS =++

)41.3(

)42.3(

)43.3(

)44.3(

)45.3(

)46.3(

2211 ZZZoc == ,)(12 scococ ZZZZ −=

39

This means that the excitation is completely independent of the scanning angle. However,

this condition has not been found to be true experimentally [21]. The use of a higher order

model should lead to a solution which can better predict scan angle blindness.

3.4.6 Automatic Tuner

Fig 3.9: A diagram indicating possible elements used in a mutual coupling corrective network

Fig 3.10: A diagram indicating a network implementation of an automatic tuning alignment system for mismatches between the power amplifier and the antenna.

40

Due to the active impedance model, any reflections caused by mismatch in impedance due to

manufacturing variations will disrupt the field pattern, reduce the overall gain of the system

and disrupt the VCO (RF power source) as depicted in Fig. 3.10. Therefore, a structure was

proposed that automatically aligns the matching network to minimize the reflected energy

that is caused by a mismatch in the impedance of the antenna to the system impedance that

appears at the antenna terminals. The automatic tuner system proposed can be described by

four parts: a matching network, a coupler, power detector, and an algorithm. The matching

network could be performed using a standard 12 +Q match that is matched for frequency

and varied by varactors. A directional coupler is used to measure the reflected energy caused

by the mismatch at the antenna terminals. The coupler that is usually built using coupled

microstrip lines can alternatively be done using discrete components as shown in Fig. 3.11.

Trades offs exist between increased component cost, size, directivity, and decreased

bandwidth.

The design procedure can be broken into two steps [31]

1. Input Impedance and Coupling Coefficients

and

2. Component Values

eoYYY 000 =

.00

00

eo

eo

YYYY

c+−

=

( ) ,2

0021

eo YYL

+=ω

)47.3(

)48.3(

)49.3(

41

and

with

The power detector used to generate an objective function measures the amount of reflected

energy. The gradient algorithm then adaptively adjusts the matching network to minimize the

objective function. A gradient algorithm was chosen based on the author’s previous

experience using optimization algorithms to design microwave networks. The algorithm

leads to a fairly robust solution, due to its continuous nature.

,1 013

21211 ωω

eYCL

C =−=

.44332211 CCCC ===

Fig 3.11: A diagram that shows a structure of a discrete version of a directional coupler [31].

ω200

13oe YYC −

=

)50.3(

)51.3(

)52.3(

42

3.4.7 Modulation Format

It will be shown that the phased array pattern is independent of a given modulation

scheme assuming no carrier frequency change [27]. For example, a QPSK modulation scheme

can be described in terms of the following excitation per symbol

for 4 ,3 ,2 ,1=i and the excitation coefficients at the antenna terminals can be represented as

Fig 3.12: A diagram showing a QPSK modulation (Left) and QAM modulation in terms of its real and imaginary components (Right).

⎥⎦⎤

⎢⎣⎡ +−+= p

op

pbaseband itAA φπω

2)1(sin )53.3(

.]

2)1[(

mod

poij

p

pRF eAAA

φπ+−

= )54.3(

43

The field pattern is then related to the element excitation by Eq. 3.38. Inserting the excitation

into an active field pattern, and after factoring the common term,]

2)1[(

mod

π−ij

eA , from the

excitations, the field pattern can be written as

which is independent of the modulation angle. The ijS are scattering parameters from port i to

port j. This can be generalized to any modulation scheme. The architecture presented is more

suitable for QAM and QPSK modulations shown in Fig. 3.12. The bit rates of the system will

be limited by the convergence of the phased locked loop, which is dependent on the loop’s gain

and bandwidth. There exists a tradeoff between the convergence of the loop and the amount of

phase noise that is introduced by the phased locked loop due to the bandwidth of the loop. A

phase aid can be used to mitigate the tradeoff between convergence and noise. Bruce Fu (a

colleague working on the system) will be continuing to work with phase aiding in his thesis.

,)(

)( ]2

)1[(

modsin)2/(2112222

sin)2/(1221111

π

φ

φ

φ

−

− ⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

+++

++∝

ij

kdj

kdj

element eAeSASAA

eSASAAEE )55.3(

44

3.4.8 Polarization Adjustment

Typically, MIMO configurations include the use of orthogonal polarization antennas

to increase the diversity gain of the system and the use of switches [38-40]. This type of

scheme does increase the probability of receiving a transmitted signal; however, a maximum

3 dB loss can occur over all available polarization of the received signal using such a

diversity technique. Additional antennas can be added at arbitrary polarizations to overcome

the loss, but hardware cost increases as a function of the number of additional antennas.

Three different types of polarizations (linear, circular, and elliptical) can be excited by a

microstrip antenna shown in Fig. 3.13. Any wave can be broken down into right handed

waves and left handed waves [41]. This can be demonstrated using a linear polarized field

[41]

Fig 3.13: A diagram showing a microstrip patch antenna configuration in terms of a measurement axis.

45

Assume that the wave can be broken down into left handed and right handed circular

polarized waves represented as

A right handed wave takes the following form:

A left handed wave takes the following form :

Substituting, these into Eq. 3.38 gives

These procedures can be generalized to any mixture of polarizations by properly weighting a

linear combination of right and left handed polarized waves. The time average power

received can then be represented as

where TRθ is the angle in between the receiving and transmit vector as shown in Fig. 3.14.

The orthonormality of left handed and right handed polarized waves allows one to form a

basis that an auto-alignment system can be built [42]. The maximum misalignment occurs at

( ) .)ˆˆ(2

jkxzy

orc eajaExE −−=

( ) .)ˆˆ(2

jkxzy

olc eajaExE −+=

( ) .ˆ).()( jkxoylcrc eEaxExExE −=+=

( ) .ˆ jkxoy eEaxE −=

),(cosˆˆ 22TRRxTxreceived aaP θ=•∝

( ) ).()( xExExE lcrc +=

)56.3(

)57.3(

)58.3(

)61.3(

)60.3(

)59.3(

46

90 degrees. This is referred to as an orthogonal polarization and results in zero power

reception. A probability distribution can be assumed as a random uniform variable with a

distribution as shown as

The SNR is a function of the polarization alignment

assuming a zero mean signal, and independence between the polarization of the signal and

the signals content. The average degradation factor of the signal-to-noise ratio, SNR, due to

misalignment can then be calculated by taking the expectation value of the SNR

Fig 3.14: Polarization unit vectors of the transmitter and receiver in terms of the polarization loss factor. [14]

.2

02)(⎩⎨⎧ <<=

πθπ

θ TRTRf

[ ][ ] ( )[ ]],([cos

)()( 2

2

2

2

2

tEtnEtxESNR TR

noise

message θσσ

==

,41

162)(cos

2max

22/2

max SNRdSNRo

TRTRTR

⎥⎦

⎤⎢⎣

⎡−== ∫

ππ

θθπθ

γπ

)62.3(

)63.3(

)64.3(

47

where

Similarly, channel capacity can be defined as

assuming a constant signal-to-noise ratio. Treating the SNR as a random variable, the

average degradation of the channel capacity can be represented as

Multi-path fading occurs due to reflections from objects and consists of a dominate line-of-

sight (LOS) component and a diffuse non-line-of-sight (NLOS) component [43]. The NLOS

component’s polarization changes from the dominate LOS component due to reflections

from obstacles [44]. Therefore, the received signal can be represented as a summation of

varying polarized NLOS components and a dominate LOS component as was shown in Fig.

1.1. An automatic alignment scheme can be implemented by adding extra bits in the

preamble of the transmitted data. The bits either represent that the signal strength has

increased or decreased. Searching of the alignment space can be done using the binary chop

algorithm with a maximum computational time varying as log(N) or through a linear search

algorithm with a maximum computation time depending on N where N is the number of

divisions in between 0 and 90 degrees. The advantage of the binary chop algorithm, shown in

Fig. 3.15, is that half the search space is discarded with each estimate of the signal. A

.)](cos1[log22/2

max2∫ +=π

θθπθ

oTRTR

TR dSNRC

],1[log2 SNRC +=

.2

2

maxnoise

messageSNRσσ

= )65.3(

)66.3(

)67.3(

48

disadvantage of the algorithm is the fact that multi-path fading or white noise can cause the

algorithm to improperly discard the interval containing the correct polarization. With the

linear search, shown in Fig. 3.16, computational time is increased but performance under

multi-path fading and noisy environments should be more robust. In order to achieve zero

loss due to polarization misalignment, the antenna’s polarization vectors must be exactly

aligned or in parallel. This can be accomplished by weighting the signals into the feeds of the

antenna.

Fig 3.15: A flow diagram of a binary search algorithm.

49

The field can then be represented by

),ˆˆ()ˆˆ( 21 zyfeedzyfeedtotal ajaAajaAE −++∝

which will produce a right handed,

and a left handed,

circularly polarized wave weighted by the excitation feed, assuming the architecture shown

on the right in Fig. 3.17 and

Fig 3.16: A flow diagram of a linear search algorithm.

.zy EE =

),ˆˆ(2 zyfeeddleftthande ajaAE −∝

)68.3(

)69.3(

)70.3(

)71.3(

),ˆˆ(1 zyfeeddrighthande ajaAE +∝

50

This assumption will occur under symmetrical conditions. The alignment can be

accomplished by adjusting the weights and phases of the signals shown on the left in

Fig. 3.17 and given by

where 1feedA and 2feedA are the complex feed excitations.

.ˆˆ 21 zfeedyfeedtotal aAaAE +∝

Fig 3.17: Dual polarization excitation structures for microstrip antenna.

)72.3(

51

3.5 Summary

All previous results were judged on implementation, cost effectiveness, and

manufacturability following current and future industrial trends. The author acknowledges

that future trends may change based on innovation but remains firm on his criteria. A novel

phase shifter using a shift register and phased locked loop was shown which has advantages

over traditional techniques in cost, repeatability, linearity, and resolution capabilities. The

disadvantage of the structure is that the resolution is limited by a master clock and sensitivity

of the digital division used in most PLLs. An offset PLL was then introduced to remedy this

sensitivity issue. The mutual coupling theory was derived from Pozar’s work [13] and show

that the baseband pre-distortion techniques can theoretically correct for mutual coupling. The

pre-distortion technique will only enhance the results of hardware mutual reduction methods.

An inconsistency was also shown with scan blindness and the current theory of mutual

coupling. A benefit of a baseband phase control methodology is that it is guaranteed not to

disturb the fields of the antenna that are based on the boundary conditions surrounding the

antenna. With the addition of more hardware, it may be possible to compute the pre-

distortion phases dynamically to mitigate the effect of external reflections. A new

measurement technique for measuring coupling parameters based on an optimization routine

on the far field pattern was introduced which will reduce testing time in large arrays. An

adjustable amplifier was also shown with an advantage over previous structures based on the

fact that zero phase distortion is possible. Phase aid using charge injection reduces

complications of stability of the phased locked loop, charge injection caused by switches, and

reduces the filters dependency on the lock detect. The automatic tuner derived in this

52

dissertation and found in previous work is very simple; the works were however developed

independently [28]. The primary difference includes the type of coupler, tuning element and

algorithm used. In terms of the coupler used, there will be a tradeoff between cost,

directivity, and bandwidth. A more detailed analysis of the difference of the tuning element

such as linearity, convergence and tuning range will be left for future work. The analysis of

the modulation types shows that the platform is capable of supporting a novel result. The

polarization method using excitation of different modes within the patch is more elegant

compared to other methods based on cost factors for adding additional antennas in a diversity

scheme and it will not induce harmonics such as a varactor being used directly on the antenna

would. Overall, new theoretical ideas were presented which will only enhance the field of the

antenna design through cost savings, better accuracy, and new methodologies.

53

CHAPTER 4 EXPERIMENTAL SETUP

4.1 Introduction

An experimental setup was implemented to expedite and increase the accuracy of the

previous antenna setup at Iowa State University. An automated network using a coherent

network analyzer and a non-coherent spectrum analyzer in combination with an optical

detector is presented. This will increase the accuracy of the measurements and expedite the

measurement results. Filters and a measurement setup are presented to allow for future

continuation and repeatability of the work presented in this dissertation.

4.2 Setup

In the initial experimental setup, a two element dipole and a microstrip antenna arrays

were used to work out initial flaws and to verify theoretical results that were previously

published.

Fig 4.1: An automatic test setup for non-coherent measurement of an array

54

This antenna setup is shown in Fig. 4.1 and Fig. 4.2, and was later expanded to the 3x3 phased

array. Fig 4.1 allows for a non-coherent detection whereas Fig 4.2 allows for a coherent

measurement system. The relevancy of coherency is determined by whether one needs carrier

phase information from the system. As proposed by Pozar [13], the mutual coupling

parameters can be gathered solely from the far field measurements which require phase

information or coherency of the wave front at various locations in space. For the noncoherent

system, the spectrum analyzer is connected to a computer which synchronizes the instruments

to an angular rotary device. The spectrum analyzer is configured for narrow band

measurements that are averaged to reduce measurement variation by the square root of the

average factor. The reduction in variation allows for low side lobe measurements to be

performed. The exact phase differences, 12 φφ − , and amplitudes, 1C and 2C , between the input

signals were measured using an oscilloscope and these signals can be described by the

equations below:

and

2 2 2( ) sin( ).rfV t C tω φ= +

Fig 4.2: An automatic test setup for a coherent test setup.

1 1 1( ) sin( ),rfV t C tω φ= + )1.4(

)2.4(

55

Scattering parameter measurements of the amplifier, filter and interconnecting cables were

combined to obtain

where

and

for 2 ,1=p using the properties of a linear system. The scattering parameters of the

array are directly measured and combined with the active field pattern to predict field

pattern measurements. The phase shifts shown were originally performed using

transmission lines, shown in Fig. 4.3, for time delays. These were later replaced by

phased locked loops as phase shifters.

, , ,( ) sin( )p antenna p antenna rf p antennaV t C tω φ= +

Fig 4.3: A phase shifter using microstrip delay lines

, , , , 0p

p antenna p cable p filter p ampφ φ φ φ φ= + + +

)3.4(

)4.4(

)5.4(

pcablepamppfilterpantennap AGGGC ,,,, =

56

The delay lines were formulated to provide an incremental phase shift between 0

and 180 degrees. A resistive splitter was designed and is shown in Fig. 4.4. The resistive

architecture was chosen over the lower insertion loss architecture primarily due to the

compactness of the design. The purpose of the splitter is to provide uniform phase and

amplitude excitations to both of the elements. Harmonics at multiples of the VCO’s

output frequency will be generated in the phased locked loop and power amplifiers due

to nonlinear effects of the transistors, and saturation effects from the oscillation loop

gain of the VCO being greater then one. The harmonics may in turn distort the field

pattern of the array due to the fact that antennas, depending on their type, are typically

Fig 4.4: A resistive splitter for equal amplitude and phase excitation of the array

57

resonant at multiples of the fundamental design resonance. These harmonics are of particular

concern compared to other spurious emissions due to the fact that patch antennas will

resonate at multiples of the design frequency. In order to mitigate the harmonic effects, a low

pass filter was placed in front of the phased locked loop with the pass band containing the

fundamental and the stop band containing the harmonics. A low pass architecture shown in

Fig. 4.5 was chosen over band pass architecture due to the fact that transmission line based

band pass filters have a tendency to resonate at multiples of the design frequencies. The

transmission line designs were chosen based on the ease of manufacturing, debugging

purposes, and good agreement between experimental and simulation results. The measured

results are shown in Fig. 4.6.

Fig 4.5: A stub low pass filter used for harmonic removal

58

The insertion loss of the filter is negligible for the purpose of this study and has 40 dB

of attenuation at the 2nd harmonic frequency. A band pass filter shown in Fig. 4.7 was placed

Fig 4.6: Stub low pass filter’s magnitude response

Fig 4.7: A band pass filter

59

behind the spectrum analyzer in order to limit thermal noise content and response to

interfering signals that were received by the horn antenna. The measured results are shown in

Fig. 4.8. Note that the filter resonates at the fundamental design frequency and at a multiple

of the primary resonance. At higher frequencies, the filter has a high pass filtering response.

This can be explained by increased coupling and surface wave effects.

Fig 4.9: Coupled line filters

Fig 4.8: The measured response of the bandpass filter

60

Additional couple line filters, as shown in Fig. 4.9, were fabricated and tested. However, the

author was not satisfied with their experimental results and they were thus discarded from the

design. Power amplifiers chosen for the design were Mini-Circuits® lee_39. The amplifiers

where chosen based on their relatively high 1 dB compression point of 18 dBm and their

broad band gain up to 10 GHz. They can be used in other applications such as for an

automated tester for an anechoic chamber. The lab setup uses a receiving horn antenna with

about 28 dB of gain. This allows for a high degree of measurement accuracy for the

transmitting antenna. The rotary motor for the transmitting antenna was then synchronized

with the spectrum and network analyzer to make automated power measurements. The

distance between the transmitting and receiving antennas was such that far field pattern

measurements were taken. The lab setup is shown in Fig. 4.10. Optical cables, such as shown

Fig 4.10: Automatic test setup consisting of horn antennae in the (Upper Left), array in the (Upper Right), synchronous step rotary (Lower Left), and spectrum analyzer (Lower Right).

61

in Fig. 4.11, were used with a bar code strip to synchronize the rotary motor and the

measurements taken with the network analyzer. The critical angle of propagation for the

optical decoder to determine the positioning of the cables as shown in Fig. 4.11 is [46]

where 1n and 2n are the indices of refraction of the core and cladding respectively and

21 nn ≥ . This can be related to the acceptance angle, aφ , by [46]

Fig 4.11: Optically encoded turn where the black marker indicates zero and white marker indicates a one [46].

,1sin2

1

21⎟⎟⎠

⎞⎜⎜⎝

⎛−= −

nn

cα

( ) ( ),sinsin 1 caa nn αφ =

)6.4(

)7.4(

62

where an is the index of refraction of the air. The cables were experimentally positioned so

that the counted number of serial bars were repeatable at various speeds of automation. A

measurement was taken by the spectrum analyzer every time a white marker was detected.

The measurements could then be correlated with the relative position of the markers.

4.3 Summary

The test setup was judged on the ability to repeat the experimental results, accuracy

of the measured data, and the reduction of testing time (relative to the previous laboratory

setup). The resolution of the automatic test setup using an optical encoder allowed for

measurements of less than a degree to be taken, which is better than the spatial resolution of

the horn antenna. To reduce the noise of the system, the antenna was rotated and the

measurement results were averaged. This led to results that were measurable up to the

accuracy of the spectrum analyzer, which is approximately 1/6 dB. In summary, the test

setup allowed for repeatable and accurate measurements, only limited by the accuracy of the

test equipment itself. Overall, the measurement system was a success and allowed for rapid

accurate measurements.

63

CHAPTER 5 EXPERIMENTAL AND SIMULATION RESULTS

5.1 Introduction

Experimental results are presented for the phase shifters using a scaled version of the

hardware phase shifter. The accuracy of the active impedance model for a 2x1 microstrip

array and a dipole array is compared against experimental results, the standard model with

and without mutual coupling effects. Steering was demonstrated in the 2x1 dipole array and a

3x3 microstrip array using the hardware phase shifter units. Simulation results of the auto-

tuner and the polarization alignment scheme are presented that allow for a more optimal

system.

5.2 Phase Shifter

5.2.1 Time Delay

The results of the waveform shifted at baseband versus the RF phase shift are shown

in Fig. 5.1. The correlation between baseband and RF phase shift was stated in Eq. 3.3.

Fig 5.1: A baseband phase shift at 3.125 MHz (Left) and an RF phase shift at approximately 2.7 GHz (Right).

64

Original measurements were made for the phase shift without scaling taken into account and

thus the waveform at RF and baseband differed by the divide ratio multiplication. Fig. 5.2

shows the results of the percent error with the multiplication factor taken into account. The

accuracy limited to phase noise reduction, which was reduced using ensemble averaging

techniques. However, there was an anomalous point which the author assumes is a cycle slip

in the d-flip flop, due to unequal time delays in the multiplexer paths.

After the correction was made to remove the multiplication factor from Eq. 3.3, the

results shown in Fig. 5.3 appear to be very exact, except at 2 major discrepancy points that

again the author assumes that there were cycle slips. This would account for the symmetry of

the points in the graph. In conclusion the results appear to match the theoretical analysis

accurately, with a minimal contamination due to phase noise error and cycle slips in the

logic.

Fig 5.2: Percentage of error for theoretical versus experimental phase shift relative to 360 degrees for an RF at 2.425 GHz, a baseband at 3.125 MHz, and phase detector at 312.5 KHz.

65

5.3 Phased Locked Loops

5.3.1 Offset Mixers

Fig 5.3: Theoretical versus experimental phase shift for the phase shifter with an RF at 2.53125 GHz, a baseband at 3.125 MHz and with a phase detector at 312.5 KHz.

Fig 5.4: A fabricated phased locked loop using offset mixers for frequency conversion of the RF down to the reference frequency.

66

A phased locked loop with offset mixers was successfully fabricated and is shown in

Fig. 5.4. The output lock frequency is shown in Fig. 5.5. The locking range of the device

was approximately 2 MHz. A coherent source needs to be utilized in order to verify that there

is no phase scaling from the input reference to the output reference. The device was not

integrated into the phased array due to the number of discrete components needed and noise

issues. Implications of this structure should be investigated in the future in terms of phase

tracking.

5.4 Antenna and Arrays

5.4.1 Antenna

The simulated directivity and the gain of a single patch element used are listed in

Table 5.1. This is typical of a microstrip patch antenna on a 30 mil board with a dielectric

constant of 3.38.

Fig 5.5: Output of the phased locked loop using offset mixers with a RF at 2.453125 GHz, and baseband at 3.125 MHz

67

Better improvement in gain and directivity can be improved by better impedance matching

and by varying the board thickness respectively. Fig. 5.6 shows a HFSS simulated pattern of

the patch.

Fig 5.6: The electric field strength emanated by a microstrip patch antenna at 2.425 GHz [45].

Table 5.1: HFSS simulated gain and directivity of microstrip patch at 2.425 GHz [45].

68

5.4.2 Array Design

A 3x3 phased array was fabricated given the previous architecture and a picture of the

array is shown in Fig. 5.7 as it was installed in the anechoic chamber. In order to calibrate the

system output phases and amplitudes of each element were measured by removing the

antenna array, which allowed direct measurement of the outputs of the individual VCO’s.

Fig 5.7: A functional 3x3 phased array.

69

It will likely be necessary to integrate the phased array onto a single PCB board to minimize

coupling effects. However, sufficient control of the experimental phased array was

accomplished to demonstrate steering of the array. The coupling between wires in the

external wiring, shown in Fig. 5.8, affects the phases on each antenna. Any slight capacitive

delay or cross coupling effect at baseband gets multiplied by the scaling factor. If sufficient

phase shifting resolution were obtained by the clock, then a possible solution would be to

calibrate the system and use pre-distortion to account for variations in wiring. The results of a

two by one dipole array correspond with Eq. 2.16. as demonstrated by measurements shown

in Fig. 5.9. The discrepancy in shape can be partially explained by finite ground plane effect,

or edge effect and other distortions such as wobble of the measurement system turn wheel.

The measured results should in theory be able to be extrapolated up for larger arrays, where

mutual coupling will become more prominent.

Fig 5.8: Phase shift circuitry

70

The results indicate that a large array can be completely controlled by a phased locked loop.

The measured phase shift for the 3x3 antenna is shown in Fig. 5.10. This demonstrates the

ability to perform phase steering with the current circuitry. The circuitry would need to be

improved before a correlation between the experimental and active impedance model and

pre-distortion technique could be better verified.

Fig 5.9: Theoretical verses experimental field patterns of a 2x1 dipole array operating with RF at 2.425 GHz and baseband at 3.125 MHz for two different phase shift patterns.

Fig 5.10: The 3x3 phased array measurements with an RF at 2.53125 GHz, a baseband at 3.125 MHz and a phase detector at 312.5 KHz. Two different phase shifts were demonstrated.

71

5.4.3 Automatic Tuner

Fig. 5.11 shows the reflection coefficient, 11S and 22S , for two mutually coupled

antennas depicted in Fig 3.10 that are tuned to a resonant frequency through the use of the

steepest decent algorithm. The antennas were originally designed for resonant frequency of

2.425 GHz. The capacitance in the 12 +Q match were varied to minimize the error (reflected

energy) that is caused by a mismatch in the impedance of the antenna. Using a typical

varactor tuning range as a constraint and 15 dB of directivity on each coupler, the auto-tuner

was able to adaptively tune over a 200 MHz bandwidth. This would well meet the demands

of typical misalignment due to manufacturing errors and also meet most frequency hopping

applications requirements. When using varactors, placement of the varactors should be made

with extreme caution. Careful placement will minimize the harmonics generated by varactors

Fig 5.11: A diagram showing the tuning capabilities of the automatic tuner for frequencies outside the resonant frequency for two microstrip patches that are coupled.

72

due to high output power coming from the antenna. Other more suitable techniques may

exist. Other practical issues that may exist include the effects of interfering received signals,

fixed point operation of a DSP or FPGA, thermal noise, and energy reflected back towards

the antenna by external objects. A coupler was successfully fabricated and is shown on the

left in Fig. 5.12. The measured results of the coupler are shown on the right in Fig. 5.12.

These results meet the bandwidth requirement of 200 MHz with a directivity of 14 dB.

5.4.4 Polarization Adjustment

Simulations results, as shown in upper left of Fig. 5.13, show that the linear search

used takes longer to converge as compared to the binary chop algorithm used under a worst

case scenario when the polarization of the transmitting and receiving antennas are

orthogonal. The faster convergence of the binary chop algorithm is due to the fact that half

the search space is discarded with each sample point. The lower of Fig. 5.13 shows the

performance of both algorithms under different SNR environments. The linear search

algorithm performs worse under extreme noisy conditions as compared to the binary chop

Fig 5.12: A fabricated version of discrete coupler (Left) and the measured coupling and insertion loss of the coupler (Right).

73

Fig 5.13: The figures show the convergence results of binary chop and a linear search algorithm (Upper Left). The figure also contains the convergence of algorithms using a multipath, 20 uniform scatterer environment (Upper Right). The results for different SNR environment for both algorithms (Lower).

74

algorithm. This can be attributed to the fact that under extreme noisy environments, the

binary chop algorithm has an equal probability of discarding the correct interval or choosing

the incorrect interval. However, the linear search algorithm performs better at a higher SNR

due to the fact that the binary chop algorithm has the possibility of discarding the wrong

interval while the linear search algorithm will result in an approximate region of the correct

polarization. The upper right diagram of Fig. 5.13 shows similar trends as does the lower of

Fig. 5.13 that was simulated under a multi-path environment. The multi-path environment

was approximated by twenty scatterers using a random uniform variable for the polarization

component.

5.4.5 Mutual Coupling

Fig. 5.14 shows a difference between the experimental results and the standard dipole

array model. Improvements are found when the active impedance, Eq. 2.16, is taken into

account. The most noticeable differences in the field pattern can be seen at lower levels of

the field pattern. This can be attributed to the domination of the coupling parameter terms

Fig 5.14: Experimental (Exp) versus theoretical results using the standard model (Std) and the active impedance model (Act) for a 2x1 dipole array at 2.425 GHz

75

122111 SASA + and 211222 SASA + ,

where 1A and 2A are the complex excitation coefficients of the array. The main beam seems

to be only mildly affected by the coupling parameters. From the experimental data of a

microstrip array shown in Fig. 5.15, there is a slight improvement between the experimental

and active impedance models. This can be attributed to smaller coupling parameters. The

active impedance model may also be viewed as a result of a first order approximation of a

Taylor expansion

with respect to the excitation coefficients. Since there are nonlinear effects in the system, the

use of a higher order model may produce a more accurate model of the system. The

coefficients can be found by using an optimization routine to minimize the error.

Fig 5.15: Experimental (Exp) versus theoretical results using the standard model (Std) and the active impedance model (Act) for a 2x1 microstrip array at 2.425 GHz.

,),(),( sin)2/(212

sin)2/(211

φφ kdjkdj eAAfeAAfE −+∝ )2.5(

)1.5(

76

5.4.6 Mutual Coupling Optimization

The results of an optimization, Eq. 3.39, can be seen in Fig. 5.16. There is an

improvement over the results in Fig. 5.15 in terms of the low level field patterns.

Minimization should be performed at multiple scanning angles in order to verify a correct

solution.

5.5 Summary

The experimental results were judged on its correlation to the theory described in

chapter 3, as well as whether deviations from the theory could be explained. New ideas were

highlighted for theoretical development. Experimental results for the phase shifter followed

very close to the theoretical development, both when scaling was present and when it was not

present. Discrepancies that were presence were attributed to hardware implementation issues.

Using the phased locked loop as phase shifters, the pattern was steered. The automatic tuner

simulation

Fig 5.16: Dipole Array with length of 1203 mils spaced 1155 mils apart, showing a comparison between active impedance model (Act), standard (Std), and optimized (Opt) results at 2.425 GHz.

77

shows that the antenna is capable of being tuned at different frequencies. Future work needs

to be done in experimentally proving the automatic tuner result. Using Matlab and its

mathematical libraries, the idea of using a dual fed microstrip antenna to automatically align

the polarization of receiving and transmitting signals was demonstrated both in the presence

of white noise and in a multipath environment. The experimental trends for the field pattern

of the array follow the theoretical model to a great deal of accuracy for the dipole array, but

showed same deviation from theory for the microstrip array. The inaccuracies of a microstrip

experiment can be explained by the existence of higher order modes present in the patch, or

due to the presence of a finite ground plane. Mutual coupling parameters were

experimentally shown to be obtained through optimization routines of the far field

measurement. The final array design demonstrates that the beam is capable of being steered

with phased locked loops. Further work needs to be performed in order to perfect the

circuitry and the system.

78

CHAPTER 6 FUTURE WORK, SUMMARY, AND DISCUSSION

6.1 Introduction

The project demonstrates antenna beams can be digitally steered using a FPGA and a

phased locked loop to control the phased array. New experimental techniques were

developed to allow more precise characterization of the system, which allows for more

accurate prediction of the beam. New items for investigation, such as scan blindness, scaling

of the input phase, phase noise, impedance variations, and mutual coupling effects in external

circuitry, have been shown to cause a deviation from the current theory. The next step for this

project would be to develop an integrated circuit using a larger phased array in order to be

able to control both the main beam and null locations. Additional techniques were

theoretically developed to increase power gain and to resolve manufacturability concerns.

New architectures were presented to resolve the scaling factor of the input phase, allowing

for a fast hopping phase system. Future work needs to be performed in these areas before a

successful implementation of the system into a practical system can be done.

6.2 Overview of Successes

Through experimental data, the system using shift registers and phased locked loops

was shown to be able to control the phase and thus the beam of the main antenna system. The

theory of an active impedance model was experimentally shown to be verified and a

development of a mutual coupling corrective network was developed. Methods for auto-

79

tuning, amplitude adjustment without phase distortion and auto-polarization alignment were

discussed and/or simulated.

6.3 Future Work

Future work should include experimentally building and verifying the method for

auto-tuning, amplitude adjustment without phase distortion, and the auto-polarization system.

Future work should also theoretically further the active impedance model and scan blindness

effects. It is suggested that one build a system on a single PCB board and use computational

techniques to further enhance the theoretical versus experimental results.

6.4 Conclusion

Overall, the project had a lot of setbacks and successes. The author personally

believes technology needs to mature and become less expensive before integrating into

commercial technology and thus mitigating the issues that have cropped up in the external

circuitry. At best, the project has laid a solid foundation for both theoretical and

experimental analysis.

80

BIBLIOGRAPHY [1] E. Lawyer, The suitability of OFDM as a modulation technique for wireless

telecommunications, with a CDMA comparison, Ph.D. diss., James Cook University,

2001.

[2] D.S. Goshi, K.M. Leong, and T. Itoh, “A secure high-speed retrodirective

communication link,” IEEE Transactions on Microwave Theory and Technique, vol. 53,

no. 11, pp. 3548-3556, Nov. 2005.

[3] R. Huang and Y. Manoli, “Phased Array and Adaptive Antenna Transceivers in

Wireless Sensor Networks,” Euromicro Symposium on Digital System Design, Rennes,

France, pp. 587-592, Aug. 2004.

[4] D. Lee and T.N Wai, “Beamforming system for 3g and 4g wireless LAN

applications,” European Conference on Circuit Theory and Design, Cork, Ireland, vol. 3,

pp. 137-140, Aug. 2005.

[5] J.B. Schoebel, T. Reimann, M. Ulm, M. Schneider, M. Jourdain, A. Carchon, an

G.J. Tilmans, “Design Consideration and technology assessment of phased-array

antenna system with RF MEMS for automotive radar applications,” IEEE Transactions

on Microwave Theory and Technique, vol. 53, no. 6, pp. 1968-1975, Jun. 2005.

[6] A. Mohamed, Ardtool, http:///www.mathworks.com/matlabcentral/fileexchange/

loadFile.do?objectId=2835&objectType=file

81

[7] M.E. Shaban, Smart Antenna System for Mobile Communications,

http://www.mathworks.com/matlabcentral/fileexchange/loadFile.do?objectId=19030&obj

ectType=File

[8] R.M. Gabriel, G. Tan, and J.S. Hayden, “RF Mems Phase Shifters: Design and

Applications,” IEEE Transactions on Microwave Magazine, vol. 3, no. 2, pp. 72-81, Jun.

2002.

[9] H.A. Nakano, R. Kobayashi, and R. Yamauchi, “A Patch Antenna Surrounded by

Parasitic Y Elements for Beam Scanning,” IEEE Antennas and Propagation Society

International Symposium, Albuquerque, New Mexico, pp. 2317-2320, Jul. 2006.

[10] W.G. William, J.M. Hui, M. Bell, F. Iskander, and J. J. Lee, ”Low-Cost Microstrip-

Line-Based Ferrite Phase Shifter Design for Phased Array Antenna Applications High-

gain,” IEEE Antennas and Wireless Propagation Letters, vol. 6, pp. 86-89, 2007.

[11] R.A. York and T. Itoh, “Injection and Phase-Locking Techniques for Beam Control,”

IEEE Transactions on Microwave Theory and Techniques, vol. 46, no. 11, pp. 1920 –

1928, Nov. 1998.

[12] K.C. Wan and Q. Xue, “Indirect Controlled Phased Source,” IEEE Microwave and

Wireless Components Letters, vol. 16, no. 12, pp. 702-704, Dec. 2006.

[13] D.M. Pozar, “The Active Element Pattern,” IEEE Transactions on Antennas and

Propagation, vol. 42, no. 8, pp. 1176-1178, Aug. 1994.

[14] C.A. Balanis, Antenna Theory, Analysis and Design, 3rd Ed., New York: John Wiley

and Sons, 2005.

82

[15] W. Werner, Lecture notes to Introduction to Microstrip Antennas, University

Karlsruhe, 2001.

[16] R.J. Mailloux, Phased Array Antenna Handbook, 2nd Ed., Norwood, Massachusetts:

Artech House Publishers, 1994.

[17] S.J. Orfanidis, Electromagnetic Waves and Antennas, [Online Doc.], New Jersey:

Rutgers University, [Feb. 2008]. Available at http://www.ece.rutgers.edu/~orfanidi/ewa/

[18] N.I. Dodov and P.Z Petkov, “The Surface Waves’ Impact on the Coupling Effect in

Microstrip Antennas,” Microwave Review, vol. 8, no. 1, Dec. 2002.

[19] P. R. Haddad and D. M. Pozar, “Anomalous Mutual Coupling Between Microstrip

Antennas,” IEEE Transactions Antennas and Propagation, vol. 42, no. 11, pp. 1545-

1549, Nov. 1994.

[20] D. M. Pozar “Input impedance and mutual coupling of rectangular microstrip

antennas,” IEEE Transactions on Antennas and Propagation, no. 11, vol. 30, pp. 1191-

1196, Nov. 1982.

[21] A.O. Kasyanov and V.A. Obukhovets, “Blindness angles in microstrip phased arrays

patterns,” International Conference on Mathematical Methods in Electromagnetic

Theory, Kharkov, Ukraine, vol. 1, pp. 280-282, Sept. 2000.

[22] S. Wanner, R.J. Weber, and J.M. Song, “Mutual coupling in phased array,” IEEE AP-

S International Symposium, Honolulu, Hawaii, pp. 153-156, Jun. 2007.

83

[23] H. Chiu, C. Cheng, R.D. Murch, and C.R. Rowell, “Reduction of Mutual Coupling

Between Closely-Packed Antenna Elements,” IEEE Transactions on Antennas and

Propagation, vol. 55, no. 6, pp. 1732-1738, Jun. 2007.

[24] Y. Fan and Y. Rahmat-Samii, “Microstrip antennas integrated with electromagnetic

band-gap (EBG) structures: a low mutual coupling design for array applications,” IEEE

Transactions on Antennas and Propagation, vol. 51, no. 10, pp. 2936 - 2946, Oct. 2003.

[25] L.C Liu and B.H. Li, “Fast locking scheme for PLL frequency synthesizer,”

Electronic Letters, vol. 40, no. 15, pp. 918 – 920, Jul. 2004.

[26] G. Nash, “Phased-Locked Loop Design Fundamentals,” [Online document], Analog

Devices, [2006]. Available at http://www.freescale.com/files/rf_if/doc/app_note/AN535

[27] D. Stephens, Phased-Locked Loops For Wireless Communications, 2nd Ed.,

Massachusetts: Kluwer Academic Publishers, 2002.

[28] J. De Mingo, A. Valdovinos, A. Crespo, D. Navarro, and P. Garcia, “An RF

electronically controlled impedance tuning network design and its application to an

antenna input impedance automatic matching system,” IEEE Transactions on Microwave

Theory and Techniques, vol. 52, no. 2, pp. 489 – 497, Feb. 2004.

[29] T. Korosec, P. Ritos, and M. Vidmar, “ Varactor-tuned microstrip-patch antenna with

frequency and polarisation agility,” Electronics Letters, vol. 42, no. 18, pp. 1015 – 1016,

Aug. 2006.

84

[30] S. Wanner, S. Sekar, Bruce Fu, R.J. Weber, and J.M. Song, “Phased Array System

Design and Modeling,” International Symposium on Signal Systems and Electronics,

Montreal, B.C., Canada, pp. 455-458, Jul. 2007.

[31] R.J. Weber, Introduction to Microwave Circuits, New Jersey: IEEE Press, 2001.

[32] J.L. Dawson, and T.H. Lee, “Cartesian feedback for RF power amplifier

linearization,” American Control Conference, Bostron, M.A., pp. 361-366, Jun. 2004.

[33] J. Huang, “The Finite Ground Plane Effect on the Microstrip Antenna Radiation

Patterns,” IEEE Transactions on Antennas and Propagation, vol. 31, no. 6, pp. 568-573,

Jun. 1984.

[34] M. Kobasa, P. Hakan, and T. Samir, “Theory of Single Microstrip Patch,” Syracuse

University, [Online Doc.]. Available at http://surf.syr.edu/projects/SamirArrayAntenna/

[35] S.J. Blank and M.F. Hutt, “On the empirical optimization of antenna arrays,” IEEE

Antennas and Propagation Magazine, vol. 47, no. 2, pp. 58- 67, Apr. 2005.

[36] H. Zhu, Y. Wang, X. Xu, and S. Li, “Direct amplitude optimization and quantization

for ultralow sidelobe phased arrays,” International Conference On Computational

Electromagnetics and its Applications, Beijing, China, pp. 183-186, Nov. 1999.

[37] W. Stutzman and G. Thiele, Antenna Theory and Design, 3rd Ed., New York: John

Wiley and Sons, 1998.

[38] L. Xin, H. Xiujiang, and N. Zaiping, “Equivalent Relations Between Interchannel

Coupling and Antenna Polarization Coupling in Polarization Diversity Systems,” IEEE

Transactions on Antennas and Propagation, vol. 55, no. 6, pp. 1709-1715, Jun. 2007.

85

[39] W.L. Liu, T.R. Chen, S.H. Chen, and J.S. Row, “Reconfigurable microstrip antenna

with pattern and polarization diversities,” Electronics Letters, vol. 43, no. 3, Part 2, pp.

1009-1012, Mar. 2007.

[40] T. Brown, S.R. Saunders, S. Stavrou, and M. Fiacco, “Characterization of

Polarization Diversity at the Mobile,” IEEE Transactions on Vehicular Technology, vol.

56, no. 9, pp. 2440-2447, Sept 2007.

[41] D.K. Cheng, Field and Wave Electromagnetics, 2nd Ed., Massachusetts: Addison-

Wesley Reading, 1983.

[42] C.A. Balanis, Advanced Engineering Electromagnetics, New York: John Wiley and

Sons, 1989.

[43] A.F. Molisch, Wireless Communications, New York: John Wiley and Sons, 2006.

[44] J. Jootar, J Diouris, and J.R. Zeidler, “Performance of Polarization Diversity in

Correlated Nakagami-m Fading Channels,” IEEE Transcations on Vehicular Technology,

vol. 55, no. 1, pp. 128-136, Jan. 2006.

[45] Ansoft, Ansoft-HFSS, Version 9.0, http://www.ansoft.com/products/hf

[46] D.K. Mynbaev and L.L. Scheiner, Fiber-Optic Communications Technology, New

Jersey: Prentice-Hall, 2001.

86

ACKNOWLEDGEMENTS

I would like to thanks Roger Cooperation for the generous support of the printed

circuit material used for prototypes, and Texas Instruments for their generous support of the

VLSI program at Iowa State University. In addition, I thank professors Jiming Song and

Zhangdao Wang for their extended support in helping me develop my technical skills and

Rob Stolpmann for his willingness to work overtime and to do the most tedious

measurements.